Published 6 November 2026 · Updated 14 November 2026 · By Ankur Khandelwal

AI label rules by platform

Of the 4 platforms whose own policy pages we could read and quote, only YouTube asks every creator to label realistic AI content. LinkedIn asks only for deceptive synthetic media, Snapchat points to the law where you are, and Pinterest leaves it optional. TikTok, X, Instagram and Facebook are not covered yet: their pages could not be re-checked by our checker, so we do not state their rules. For YouTube, a checklist, Google’s own examples and five worked cases are below the table. (General information, not legal advice.)

The short answer

Every cell on this page comes from the platform’s own policy or help page, with the sentences quoted and the page’s own date. Where a page is silent, the cell says “Not stated on the page”. Values confirmed 13 Nov 2026.

AI label rules, side by side

The labelling duty on each platform, what triggers it, how to disclose, what happens if you don’t, and what the label is called. Scroll sideways on a phone.

Download all 4 rows with every quoted sentence: CSV · JSON. For what the law says in each country, see AI labelling laws by country.

YouTube AI label: the disclosure checklist

YouTube is the one platform here that asks every creator to label realistic AI content. Its rule: “We require creators to disclose AI-generated or altered content that seems realistic.” “Creators can make this disclosure during the upload process.” The label follows from the “AI use” answer you give in YouTube Studio. Answer these in order for each video; the first Yes decides it.

1. Does any part of it look real, and was that part made or meaningfully changed with AI? If nothing a viewer could take for real footage or a real recording came from AI, stop: no disclosure.
2. Does that part do one of YouTube’s four things? A real person appearing to say or do something they didn’t, altered footage of a real event or place, a realistic scene that never happened, or AI music as the main focus. If yes: disclose. Part of a video counts: “This could include content that is fully or partially altered or created using AI tools.”
3. Was AI only used for a minor or behind-the-scenes job? Upscaling, sharpening, colour or lighting, beauty or effects filters, captions, the script, title or thumbnail, ideas, your own cloned voice for a dub. If that is all: no disclosure.
4. Still not sure? Disclose. YouTube says “Disclosing AI content alone won't limit a video's audience or impact its eligibility to earn money.” Not disclosing is what carries the penalties.

Source for every YouTube rule in these sections: YouTube Help, “Disclosing use of GenAI content” (no date shown on the page). Three of its sentences (the rule, the upload step and the penalty) are re-read by a script every night; Google’s example lists and steps below were copied by hand on 6 Nov 2026. The checklist, the worked examples and “where the lists stop” are our reading, not YouTube’s.

YouTube: what must be disclosed, and what is exempt

Must be disclosed

Creators must disclose GenAI content that:

• “Makes a real person appear to say or do something they didn't do.”
• “Alters footage of a real event or place.”
• “Generates a realistic scene that didn't actually occur.”
• “Creates music that's the main focus of the video.”

Google’s examples of content that needs the disclosure:

• “AI generated music”
• “AI generated extra footage of a real place, like a video of a surfer in Maui for a promotional travel video”
• “AI generated realistic videos of a match between two real professional tennis players”
• “Making it appear as if someone gave advice that they did not actually give”
• “Showing a realistic depiction of a tornado or other weather events moving toward a real city that didn't actually happen”
• “Making it appear as if hospital workers turned away sick or wounded patients”
• “Depicting a public figure stealing something they did not steal, or admitting to stealing something when they did not make that admission”
• “Making it look like a real person has been arrested or imprisoned”

Doesn’t have to be disclosed

“Creators don't need to disclose non-realistic content that's made with AI, or edits to realistic content that are minor. Minor edits are ones that are primarily aesthetic, and don't alter the content in a way that could mislead the viewer about what actually happened.”

Not realistic:

• “Someone riding a unicorn through a fantastical world”
• “Green screen used to depict someone floating in space”
• “Using an AI-generated or altered animation of a missile in a fully animated video”

Minor:

• “Applying beauty filters”
• “Color adjustment or lighting filters”
• “Special effects filters, like adding background blur or vintage effects”
• “Production assistance, like using generative AI tools to create or improve a video outline, script, thumbnail, title, or infographic”
• “Caption creation”
• “Video sharpening, upscaling or repair and voice or audio repair”
• “Idea generation”
• “Cloning one's own voice to create voice overs or dubs”
• “Gameplay footage from video games”
• “AI generating or extending a backdrop to simulate a moving car”
• “Using effects to enhance previously recorded audio”

Under each list, Google adds: “Keep in mind, the above list is not exhaustive.”

YouTube: five worked examples for AI video

Our reading of Google’s lists against the kinds of clip people make with Veo and Gemini. Each one names the line it rests on.

1. A realistic Veo clip of a real place, used as B-roll

You generate a crowded sunrise at the Gateway of India for a travel Short, then take the Veo sparkle off.

Disclose: Yes. It is AI footage of a real place made to look real. Google’s list names almost exactly this: “AI generated extra footage of a real place, like a video of a surfer in Maui for a promotional travel video”. Removing the visible logo changes nothing.

2. A photoreal clip of a person who does not exist

A Gemini or Veo clip of an invented chef cooking in an invented kitchen, realistic enough to pass for filmed footage.

Disclose: Yes. No real person or place is involved, which is why this one is easy to miss. It still falls under “Generates a realistic scene that didn't actually occur.”

3. A fantasy scene

A dragon circling a floating city, or a cat in a spacesuit on the Moon.

Disclose: No. Nobody could take it for real events. Google’s not-realistic examples include “Someone riding a unicorn through a fantastical world”. YouTube may still label it on its own (see below); in most cases you can switch that back to No.

4. Your own footage, with AI help around it

You film yourself, upscale the video to 4K, add AI captions, use AI for the script and thumbnail, and dub it into Hindi with a clone of your own voice.

Disclose: No. Each of those is on Google’s list of minor edits or production help: “Video sharpening, upscaling or repair and voice or audio repair”, “Caption creation”, “Production assistance, like using generative AI tools to create or improve a video outline, script, thumbnail, title, or infographic” and “Cloning one's own voice to create voice overs or dubs”. The list says one’s own voice; a clone of someone else’s voice is not on it.

5. A real person who appears to recommend something

An AI clip in which a well-known cricketer seems to recommend your app.

Disclose: Yes, and disclosing does not make it allowed. It matches “Makes a real person appear to say or do something they didn't do.” and “Making it appear as if someone gave advice that they did not actually give”. The label is not a permission slip: “Remember, we apply our Community Guidelines to all content on YouTube, including content that is AI generated or altered using AI.” Using a real person’s likeness can also raise legal issues outside YouTube.

Where Google’s lists stop

Google says more than once that its lists are not complete. Three cases they leave open, with how we read them:

• An AI-extended background on real footage. “AI generating or extending a backdrop to simulate a moving car” is on the minor list, but “AI generated extra footage of a real place, like a video of a surfer in Maui for a promotional travel video” needs disclosure. Our reading: if the added part could make a viewer believe something happened at a real place that didn’t, disclose.
• One realistic AI shot inside a real video. Google says: “This could include content that is fully or partially altered or created using AI tools.” Our reading: a single generated shot that passes for real is enough.
• Photoreal style, impossible subject. A lifelike rendering of something nobody could believe (a city floating in the sky) sits with the unicorn example, not the tornado one. If the impossible part is subtle enough to pass for real, treat it as realistic.

When a case stays unclear, choosing Yes costs a label; YouTube says disclosure alone does not limit audience or earnings.

YouTube: how to disclose, and when it adds the AI label itself

The setting is called “AI use”, and YouTube says it is available in YouTube Studio on a computer or mobile device. Its steps:

1. “Go to YouTube Studio.”
2. “Follow the steps to upload content.”
3. “In the Attributes section, under "AI use," tap Yes if your content meets the disclosure requirements or No if not.”
4. “Continue to select other video details.”

Where the label shows. “For AI-generated music or content that is photorealistic, a label in the video player may also appear.” “Labels may appear in the expanded description for non-music AI-generated audio or AI content that is non-photorealistic or animated.”

When YouTube adds the label itself. YouTube may apply an AI label automatically, on the player or in the expanded description, for:

• “Content made using YouTube's GenAI tools”
• “Content that contains C2PA metadata”
• “Content that our internal systems detect is AI generated or altered”

“If our systems make an error, creators will have the ability to change AI disclosure in most cases by selecting No in the AI disclosure survey under "Attributes" in YouTube Studio.” “However, content made with YouTube's AI tools, content containing C2PA metadata, or content labeled after manual review can not be adjusted.” The help page does not describe a way to appeal those three.

If you don’t disclose: “Creators who consistently choose not to disclose this information may be subject to manual application of a label, or penalties from YouTube, including removal of content or suspension from the YouTube Partner Program.”

A removed watermark changes none of this. YouTube’s test is whether the content seems realistic, not whether a generator’s logo is on it, so a realistic Veo clip needs the disclosure with or without the sparkle. Google’s invisible SynthID watermark also stays in the file when the visible mark is taken off; How Clipwipe works lists what it changes in a file and what it leaves alone.

Snapchat: do not remove the Ghost with sparkles

If you post AI video made with Veo or Gemini

• On YouTube, judge by realism, not by the logo. A realistic Veo clip of a scene that did not happen needs the AI use disclosure whether or not Google’s visible sparkle is still on it. YouTube says it may also add a label itself, from its own AI tools, C2PA metadata, or its detection systems. For Google’s full lists of what needs the label and what is exempt, with worked examples, see the YouTube AI label checklist above.
• On LinkedIn, disclose anything that could pass for real people or events. The page gives no toggle, so say it in the post. Obvious parody or satire may not need it.
• On Snapchat, check the rule where you live. Snap leaves labelling to regional requirements. The by-country law page covers the EU, India, China, South Korea, Vietnam and others.
• On Pinterest, labelling is your choice, though its detection may label an image Pin anyway. The page speaks of image Pins and does not name video. Pinterest’s updated Community Guidelines took effect 12 Nov 2026; re-read on 13 Nov 2026, they still cover synthetically generated content and still set no labelling duty.
• Removing a visible watermark changes none of this. It does not remove the invisible SynthID mark in Google’s output, and it does not change whether a platform asks you to disclose. Only remove a mark from a clip you own or have the right to edit.

Getting a clip of your own ready to post? The Veo watermark remover takes the visible mark off in your browser, and which quality and resolution to choose covers the settings for uploading it to Instagram or YouTube.

What each platform says, in its own words

Each entry gives the duty, the trigger, how to disclose, what happens if you don’t and the label’s name, then the exact sentences quoted from the platform’s pages. Where a page does not say something, the entry says “Not stated on the page” rather than filling it in from news or blogs.

YouTube

Yes, for realistic AI content: say so at upload under Attributes > AI use in YouTube Studio. Clearly unrealistic content and minor edits are exempt.

• Labelling duty: Required for realistic AI content.
• What triggers it: AI-generated or altered content that seems realistic: a real person made to say or do something they did not, altered footage of a real event or place, a realistic scene that did not occur, or music that is the main focus of the video.
• Video: Yes.
• How to disclose: During upload, in the 'AI use' setting under Attributes in YouTube Studio: choose Yes if the content meets the disclosure requirements.
• If you don’t: Creators who consistently do not disclose may get a label applied manually, or penalties including removal of content or suspension from the YouTube Partner Program. YouTube may also add an AI label automatically (its own GenAI tools, C2PA metadata, or content its systems detect as AI generated or altered).
• The label’s name: Not stated on the page.
• Note: Non-realistic content and minor edits do not need disclosure. Disclosing alone does not limit a video's audience or its eligibility to earn money.

In the platform’s own words:

YouTube Help article on disclosing AI content (support.google.com):

• “We require creators to disclose AI-generated or altered content that seems realistic.”
• “Creators can make this disclosure during the upload process.”
• “Creators who consistently choose not to disclose this information may be subject to manual application of a label, or penalties from YouTube, including removal of content or suspension from the YouTube Partner Program.”

Dates on the pages: No date shown. Quotes last re-read automatically: 1 page(s) re-read, every quoted sentence matched. Values confirmed 13 Nov 2026.

LinkedIn

Only for synthetic or doctored media that could mislead, such as a person shown saying something they did not say. Obvious parody may not need it. No toggle is described.

• Labelling duty: Required for deceptive synthetic media.
• What triggers it: Synthetic or manipulated media that depicts a person saying something they did not say, or doctored images or videos that distort real-life events; obvious parody or satire may not need explicit disclosure.
• Video: Video is named only in the 'doctored images or videos' example.
• How to disclose: Not stated on the page.
• If you don’t: Depending on severity LinkedIn may limit the visibility of the content, label it, or remove it entirely; repeated or egregious offenses can result in account restriction.
• The label’s name: Not stated on the page.
• Note: No toggle or tag is described on either page.

In the platform’s own words:

LinkedIn Professional Community Policies (linkedin.com):

• “Do not share synthetic or manipulated media that depicts a person saying something they did not say…”
• “Depending on the severity of violation, we may limit the visibility of certain content, label it, or remove it entirely.”
• “Repeated or egregious offenses can result in account restriction.”

LinkedIn Help article on fake or altered content (linkedin.com):

• “LinkedIn generally requires members to disclose the fake or altered nature of this content.”

Dates on the pages: Policy page: no date shown. Help article: 'Last updated: 2 years ago'. Quotes last re-read automatically: 2 page(s) re-read, every quoted sentence matched. Values confirmed 13 Nov 2026.

Snapchat

No global rule: Snap says some regions require a label and you must meet your region’s rules. Images made with Snap’s own AI tools may carry its Ghost-with-sparkles watermark.

• Labelling duty: Where local law requires it.
• What triggers it: No global creator labelling rule. Snap says some regions require users to clearly disclose or label AI-generated content and that users are responsible for meeting the requirements in their region.
• Video: Not stated: the help article on Snap's own AI tools covers images.
• How to disclose: Not stated on the page.
• If you don’t: Enforcement can include removing violating content, issuing warnings, limiting visibility, or terminating the account. Separately, Snap's terms say not to remove the Snap Ghost with sparkles watermark that may be added to images made with Snap's AI tools.
• The label’s name: On images made with Snap's own tools: a Snap Ghost with sparkles watermark, and a Context Card.
• Note: Labels apply to content made with Snap's tools; the help article warns images created with non-Snap products may not be labeled as AI generated.

In the platform’s own words:

Snap Community Guidelines (values.snap.com):

• “Some regions require users to clearly disclose or label AI-generated content.”
• “Our enforcement actions can include termination of an account.”

Snapchat Support article on Snap’s AI tools (help.snapchat.com):

• “…a watermark of a Snap Ghost with sparkles may be added to those images…”
• “Do not remove Snap's Ghost with sparkles watermark as it is a violation of our Terms.”
• “Images created with non-Snap products may not be labeled as AI generated.”

Dates on the pages: No date shown on either page. Quotes last re-read automatically: 2 page(s) re-read, every quoted sentence matched. Values confirmed 13 Nov 2026.

Pinterest

Optional. A Gen AI label shows on an image Pin when its owner marks it or Pinterest’s systems detect AI. No penalty for not labelling is stated.

• Labelling duty: Optional.
• What triggers it: No creator duty stated. A label appears if the content owner identifies the content as modified or generated with AI, or when Pinterest's detection systems identify that it may have been.
• Video: Not stated: the page speaks of image Pins.
• How to disclose: Not stated on this page; it points to a separate 'Gen AI labels' article.
• If you don’t: No penalty for not labelling is stated. Community Guidelines apply to synthetically generated content, and Pinterest may remove, limit or block distribution of violating content and the accounts behind it.
• The label’s name: 'Gen AI labels' (the page's own term).
• Note: Re-read on 2026-11-13, after the updated Community Guidelines took effect on 2026-11-12: they still apply to synthetically generated content and still set no AI label or disclosure duty.

In the platform’s own words:

Pinterest Help: AI at Pinterest (help.pinterest.com):

• “Pinterest will show a label on the close-up of an image Pin if the content owner identifies that the content was modified or generated with AI.”
• “Pinterest will show a label on the close-up of an image Pin when Pinterest's detection systems identify that it may have been generated or altered with AI.”

Pinterest Community Guidelines (policy.pinterest.com):

• “Our policies apply to everyone and to all types of content, including synthetically generated content.”

Dates on the pages: Help page: no date shown. Community Guidelines: 'Last updated: November 2026'. Quotes last re-read automatically: 2 page(s) re-read, every quoted sentence matched. Values confirmed 13 Nov 2026.

Platforms not covered yet

These platforms were looked for, but their own pages could not be re-checked by our checker, so nothing about their rules is stated here. A platform moves into the table once its page’s sentences can be quoted and re-read every night; until then, read its page yourself before you post.

Frequently asked questions

Download the data

The one-line answers in the entries above are our own reading of the rows and are not in the files. The licence covers our compilation and wording; the quoted sentences stay with the platforms that wrote them.

How this page is kept current

Every night a script re-reads each platform page cited here and checks that every quoted sentence is still on it, word for word. If a sentence changes or a page disappears, the check fails, the last good entry stays up, and a person re-reads the page before anything here changes. The date at the top of this page moves only when an entry or this page’s own text changes.

Spotted a rule that has changed, or a platform we should add? Email nocodetalks@gmail.com.

Not legal advice

This page reports what each platform says on its own pages. It is general information, not legal advice, and a platform can change its rules at any time: its own terms are what bind you. For what AI labelling laws say in different countries, see Is removing an AI watermark legal?; for which stock sites accept AI video, see Can you sell AI-generated video on stock sites?

About the author

Ankur Khandelwal is an independent maker who builds small, practical web tools under the name nocodetalks, Clipwipe among them.

← Clipwipe home